Abstract: Based on probabilistic distribution theory,an analytical model for truncated probabilistic distribution is presented.A new optimization model for calculating reliability index is proposed,in virtue of which the reliability analysis is carried out with truncated probabilistic distributions.In this model,it is unnecessary to normalize the non-normal distributions and to learn the information of distribution type.A technique to sample the truncated probabilistic distribution in selected partial regions is proposed,in which Monte Carlo method turns to be feasible.Example results show that the truncated probabilistic distributions are more realistic,which will improve the reliability index.The influences should not be ignored.